TIBTECH - NOVEMBER 1990 [Vol. 8]
301
Electrospray ionization mass spectrometry-a powerful new analytical tool Most biochemists and molecular biOlogists have traditionally shown little interest in mass spectrometry, believing it to be the province of physicists or chemists in the pharmaceutical industry. Apart from the acknowledged value of fast atom bombardment mass spectrometry ('FAB' MS) (see Glossary) for mass and sequence determination of small peptides, it had been thought not to offer much assistance to those working with intact proteins or other large biopolymers. In 1988, John Fenn and colleagues (Dept. of Chemical Engineering, Yale) developed an earlier idea of Dole to produce an interface suitable for transferring large, polar molecules from the liquid to the gas phase for direct mass determination 1. In Dole's technique 2, aqueous solvent containing biopolymers emerges through a hollow needle and is dispersed as a fine spray by a highvoltage electric field. This is similar in principle to the electrostatic paint sprayer. Sons in the droplets are freed of solvent by a rapid countei" current of warm gas and molecular ions are drawn into the high vacuum of a quadrupole mass analyser (see Glossary) through a fine capillary. This electrospray source (ES) produces multiply charged ions. In positive ion ESMS, lysine, arginine, histidine and any free amino-termini may carry positive charges usually through protonation, although one or more protons may be replaced by sodium or potassium ions. Serum albumin, for example, produces at least 33 readily detectable molecular ions carrying between 28 and 60 positive charges. These appear as separate peaks because the quadrupole measures the mass:charge ratio. The range of molecular mass for which ESMS instruments are suitable is only 10Da to 4kDa for ~) 1990, Elsevier Science Publishers Ltd (UK)
singly protonated species, though multiple charging of larger protefns brings their mass : charge ratio within range. The mass of proteins the size of serum albumin (66kDa) and greater can be measured accurately in this way 3. The number of positive charges on each ion is worked out simply by measuring the spacing between adjacent peaks, which differ by one positive charge and usually by unit mass only. Each ion then allows an estimation of the molecular mass to be made, which can be used to improve the statistical quality of the overall determination. In the example (Fig. la), the mass determined for recombinant human albumin produced by yeast is 66440.3 + 2Da. This differs by 3 Da from the mass calculated using the cDNA-predicted amino-acid sequence (66437 Da), an error of only 0.004%. The value of mass spectrometry for solving many of the biochemical
problems associated with natural and recombinant proteins using FAB MS on small fragments was previously described in TIBTECH in an excellent review by Howard Morris and Fiona Greer 4. The electrospray source now extends these analyses to large fragments and in many cases, to intact proteins. Post-translational modifications (such as Nacetylation of the amino-terminus) can be recognized and often identified by mass difference, without fragmentation of the protein. In Fig. 2, ESMS detected the unexpected presence of an exogenous cysteine molecule linked to the single free thiol group in a recombinant 22 kDa protein. The difference between the calculated and experimental mass values was 119Da; exactly as expected for thiol-linked cysteine. This was confirmed by mild reduction upon which the higher-mass molecular ion series disappeared. The majority of human proteins of therapeutic value are glycosylated and may have further modifications such as the y-carboxyl glutamyl groups of the vitamin K-dependent blood coagulation proteins or the ~erythrohydroxyaspartic acid of protein C. Although a single ESMS run will not completely characterize these modifications, it can confirm their presence, or absence, and simplify the detailed analysis of these awkward groups. It is now a simple matter to determine whether terminal amino acids
--Glossary FABMS - (Fast atom bombardment mass spectrometry) Neutral atoms collide with sample in a liquid matrix and eject biopolymer ions. Useful range; up to 10kDa. LSIMS - (Liquid secondary ion mass spectrometry) As above but energetic particle is, for example, Cesium ion. LOMS - (Laser desorption mass spectrometry) Coherent photons eject biopolymer from a liquid matrix. PDMS - (Plasma desorption mass spectrometry) The 'plasma' is generated from californium.252 fission products colliding with a biopolymer adsorbed onto a nitrocellulose film. ESMS - (Electrospray mass spectrometry) High voltage field dispersion of charged fluid droplets - protein ions desorb from fluid surface. 'lonspray' and 'Thermospray' are variants where gas and heat, respectively, assist the ion evaporation process. QuadrupoleA mass analyser where electrostatic deflection is used to select masses reaching the detector. Time-of-flight - A mass analyser which imparts acceleration to ions, then allows them to drift freely until reaching the detector: the mass is proportional to the time-of-flight. Daughter i o n - Molecular fragment ion separated by mass/charge in a second analyser.
0167 - 9430/90/$2.00
302
TIBTECH - NOVEMBER 1990 [Vol. 8]
~Fig. 1
l°°1 have been lost through exopeptidase action. This has always been difficult to achieve satisfactorily at the carboxy-terminal end of proteins. With ESMS, 'ragged ends' can be detected down to a level of 1% for recombinant proteins of moderate mass. Protein chemists have become very familiar with the use of the peptide 'map', a digest of a protein with a protease such as trypsin, for characterization or fingerprint analysis. The peptides are separated by HPLC or the more recent, complementary technique of capillary zone electrophoresis (CZE). Neither technique guarantees resolution of all the peptides and their identities remain unknown until they are sequenced for a few residues. However, if the HPLC or CZE is coupled to ESMS then co-eluting peptides are resolved and the accurate mass determination permits identification of all the peptides (when the protein sequence is known). Although in principle, the HPLC or CZE could be dispensed with, in practice, multiple charging of peptides reduces the resolution possible. However, tryptic peptides will give a small number of positive ions since they generally contain only one lysine or arginine residue. Where the protein sequence is unknown, ESMS instrumentation offers the possibility of collisions with the atoms of an inert gas. This fragments just one selected molecular ion, with preferential breakage at the polypeptide amide bonds. The resulting 'daughter' ion spectrum can be interpreted to give the peptide sequence. As the 'footprint' of mass spectrometers becomes smaller and biochemists more aware of their analytical power, mass detectors on HPLC and CZE systems will not be an uncommon sight! Another important and unexpected benefit of ESMS is that the width of the molecular ion 'window' appears to give a good idea of protein quality. As proteins age in vivo or in vitro, sulphur-containing and aromatic amino acids oxidize, asparagine deamidates and chemical adducts may form. These changes increase both mass and heterogeneity of the product. Comparison of the peak width for human serum albumin (Fig. lb) with that of the recombinant protein (Fig. la), shows
48+~
a
80
51 -t-
40 2 _
.
,
1100
.
1200
.
.
,
.
.
.
.
.
.
.
1300
.
,
b
10(
.
.
1400
.
.
1500
1600
1700
1800
1
6( 4( 2( (
.
1100
.
.
.
,
.
1200
.
.
.
.
.
.
.
.
1300
.
.
.
.
.
.
.
1400
,
.
1500 M/z
.
.
.
.
.
.
.
.
1600
.
.
.
.
,
.
.
1700
,
.
.
.
.
1800
ESMS spectrum of (a) recombinant (1-585), and (b) native human albumin (rHA and HSA). The spectrum shows M/z (mass : charge ratio) (n is equivalent to z in this instance) values between 1100 and 1860. Each peak represents a molecular ion Mn = [M ÷ nil+] n+ where M is the molecular weight, and n is the number of positive charges (in this case protons), n can be calculated from two adjacent ions, e.g. m7 and m2. Then n = (mT-l)/(mE-m~) and M = n(m2-1). In (a) two peaks are labelled (n+). The arrow indicates that HSA has a slightly greater average mass than rHA, and it is clearly more heterogeneous [compare peak widths for (a) and (b) at 10% peak height]. (The ESMS spectra were obtained using a BIO-Q MS, courtesy of VG Biotech Ltd. The rHA was provided by DELTA Biotechnology, Nottingham.)
that even when the individual mass differences cannot be resolved, the peak width at a set height may be used as a criterion of acceptance or rejection for a therapeutic protein. The reported higher quality of recombinant human protein products, as determined by various biophysical criteria, reflects their improved homogeneity compared with their naturally derived human counterparts. The desorption process in ESMS is still not well understood. The gaseous protein ions, freed of their solvent shells, probably unfold. This is because protein folding is an entropic process, driven by the burying of hydrophobic amino acids in the protein core away from the aqueous phase. Multiple charging probably helps the opening-up process through electrostatic repulsion. Bound ligands and cofactors (such as haeme groups) are stripped off and protein subunits are separated to leave linearized apoproteins. Covalent bonds are not broken in the process and it is noticeable that proteins (such as albumin) which are rich in disulphide bonds can accept fewer charges than smaller mol-
ecules which do not form disulphide bonds. This probably results from the inability of the crosslinked polypeptide chain in these proteins to extend fully. The relatively gentle ionization and desorption in ESMS offers considerable advantages over the older techniques of FAB and plasma desorption (PDMS) (see Glossary). For these ion sources, the very high energies needed to remove protein ions from liquid and solid matrices limits the methods to masses Of 10-20 kDa and the accumulation of data is slow (minutes to hours). In addition, both methods cause fragmentation of proteins which, coupled with the interference by matrix ions, limits both mass range and accuracy. This reduces the likelihood of being able to identify posttranslational modifications without first fragmenting the protein and separating the modified peptides. Since ESMS is able to measure both high and low molecular weight species accurately, further development of ESMS for protein sequence analysis is expected. The very valuable information now obtainable quickly and simply
303
TIBTECH- NOVEMBER 1990 [Vol. 8]
~Fig. 2 lOO 95 90 85 8o 75 70 65 60 55 5o 45 40 35 3o 25 20 15 lO 5 o~ 900
19+
charged with responsibility for scrutiny of each prospective recombinant therapeutic protein submission for a product licence. Their major concerns are for the safety, purity and, in many cases, nature identity of the product. No doubt protein mass spectrometry will now have to be added to the battery of assays and analyses currently undertaken by the biotechnology industry. For the research worker, however, the new technology is likely to prove both enabling and time-saving 5. Interesting results are certain to be reported as the use of ESMS becomes more widespread.
18+
0
0
1000
1100
1200
1300
1400
1500
M/z
ESMS of single domain rHA (1-194). There are at least two series of molecular ions. Series 0 differs from series R by 119Da. The expected mass is 22233.94Da, which differs little from that determined for the R series (22232.9_+ 1Da). The mass determined for the 0 series is 22352.3_+ 1.6Da [ 0 R = 119 Da]. Reduction o f the protein with dilute thiol causes series 0 to disappear and series R to increase in abundance. This can only result from cysteine having been picked up from the culture m e d i u m by formation of a disulphide bond to a free protein thiol. -
References 1 Meng, C. K., Mann, M. and Fenn, J. B. (1988) Z. Phys. Ser. D 10, 361-368 2 Dole, M. eta]. (1968) J. Chem. Phys. 49, 2240-2249 3 Mann, M., Meng, C. K. and Fenn, J. B. (1989) Anal. Chem. 61, 1702-1708 4 Morris, H. R. and Greer, F. M. (1988) Trends Biotechno]. 6, 140-147 5 Dorsselaer, A. et al. (October 1990) Biomed. Environ. Mass Spec. (in press)
MICHAELGEISOW for proteins of relatively high molecular weight should re-awaken the interest of biochemists and biotechnologists in mass spectrometry and send them hunting through their
notes for the masses of the elements. It should also stimulate the interest of the regulatory authorities, such as the Food and Drugs Administration (FDA). The latter, of course, is
Biodigm, 115 Main Street, East Bridgeford, N o t t i n g h a m NG13 8NH, UK.
301
Electrospray ionization mass spectrometry-a powerful new analytical tool Most biochemists and molecular biOlogists have traditionally shown little interest in mass spectrometry, believing it to be the province of physicists or chemists in the pharmaceutical industry. Apart from the acknowledged value of fast atom bombardment mass spectrometry ('FAB' MS) (see Glossary) for mass and sequence determination of small peptides, it had been thought not to offer much assistance to those working with intact proteins or other large biopolymers. In 1988, John Fenn and colleagues (Dept. of Chemical Engineering, Yale) developed an earlier idea of Dole to produce an interface suitable for transferring large, polar molecules from the liquid to the gas phase for direct mass determination 1. In Dole's technique 2, aqueous solvent containing biopolymers emerges through a hollow needle and is dispersed as a fine spray by a highvoltage electric field. This is similar in principle to the electrostatic paint sprayer. Sons in the droplets are freed of solvent by a rapid countei" current of warm gas and molecular ions are drawn into the high vacuum of a quadrupole mass analyser (see Glossary) through a fine capillary. This electrospray source (ES) produces multiply charged ions. In positive ion ESMS, lysine, arginine, histidine and any free amino-termini may carry positive charges usually through protonation, although one or more protons may be replaced by sodium or potassium ions. Serum albumin, for example, produces at least 33 readily detectable molecular ions carrying between 28 and 60 positive charges. These appear as separate peaks because the quadrupole measures the mass:charge ratio. The range of molecular mass for which ESMS instruments are suitable is only 10Da to 4kDa for ~) 1990, Elsevier Science Publishers Ltd (UK)
singly protonated species, though multiple charging of larger protefns brings their mass : charge ratio within range. The mass of proteins the size of serum albumin (66kDa) and greater can be measured accurately in this way 3. The number of positive charges on each ion is worked out simply by measuring the spacing between adjacent peaks, which differ by one positive charge and usually by unit mass only. Each ion then allows an estimation of the molecular mass to be made, which can be used to improve the statistical quality of the overall determination. In the example (Fig. la), the mass determined for recombinant human albumin produced by yeast is 66440.3 + 2Da. This differs by 3 Da from the mass calculated using the cDNA-predicted amino-acid sequence (66437 Da), an error of only 0.004%. The value of mass spectrometry for solving many of the biochemical
problems associated with natural and recombinant proteins using FAB MS on small fragments was previously described in TIBTECH in an excellent review by Howard Morris and Fiona Greer 4. The electrospray source now extends these analyses to large fragments and in many cases, to intact proteins. Post-translational modifications (such as Nacetylation of the amino-terminus) can be recognized and often identified by mass difference, without fragmentation of the protein. In Fig. 2, ESMS detected the unexpected presence of an exogenous cysteine molecule linked to the single free thiol group in a recombinant 22 kDa protein. The difference between the calculated and experimental mass values was 119Da; exactly as expected for thiol-linked cysteine. This was confirmed by mild reduction upon which the higher-mass molecular ion series disappeared. The majority of human proteins of therapeutic value are glycosylated and may have further modifications such as the y-carboxyl glutamyl groups of the vitamin K-dependent blood coagulation proteins or the ~erythrohydroxyaspartic acid of protein C. Although a single ESMS run will not completely characterize these modifications, it can confirm their presence, or absence, and simplify the detailed analysis of these awkward groups. It is now a simple matter to determine whether terminal amino acids
--Glossary FABMS - (Fast atom bombardment mass spectrometry) Neutral atoms collide with sample in a liquid matrix and eject biopolymer ions. Useful range; up to 10kDa. LSIMS - (Liquid secondary ion mass spectrometry) As above but energetic particle is, for example, Cesium ion. LOMS - (Laser desorption mass spectrometry) Coherent photons eject biopolymer from a liquid matrix. PDMS - (Plasma desorption mass spectrometry) The 'plasma' is generated from californium.252 fission products colliding with a biopolymer adsorbed onto a nitrocellulose film. ESMS - (Electrospray mass spectrometry) High voltage field dispersion of charged fluid droplets - protein ions desorb from fluid surface. 'lonspray' and 'Thermospray' are variants where gas and heat, respectively, assist the ion evaporation process. QuadrupoleA mass analyser where electrostatic deflection is used to select masses reaching the detector. Time-of-flight - A mass analyser which imparts acceleration to ions, then allows them to drift freely until reaching the detector: the mass is proportional to the time-of-flight. Daughter i o n - Molecular fragment ion separated by mass/charge in a second analyser.
0167 - 9430/90/$2.00
302
TIBTECH - NOVEMBER 1990 [Vol. 8]
~Fig. 1
l°°1 have been lost through exopeptidase action. This has always been difficult to achieve satisfactorily at the carboxy-terminal end of proteins. With ESMS, 'ragged ends' can be detected down to a level of 1% for recombinant proteins of moderate mass. Protein chemists have become very familiar with the use of the peptide 'map', a digest of a protein with a protease such as trypsin, for characterization or fingerprint analysis. The peptides are separated by HPLC or the more recent, complementary technique of capillary zone electrophoresis (CZE). Neither technique guarantees resolution of all the peptides and their identities remain unknown until they are sequenced for a few residues. However, if the HPLC or CZE is coupled to ESMS then co-eluting peptides are resolved and the accurate mass determination permits identification of all the peptides (when the protein sequence is known). Although in principle, the HPLC or CZE could be dispensed with, in practice, multiple charging of peptides reduces the resolution possible. However, tryptic peptides will give a small number of positive ions since they generally contain only one lysine or arginine residue. Where the protein sequence is unknown, ESMS instrumentation offers the possibility of collisions with the atoms of an inert gas. This fragments just one selected molecular ion, with preferential breakage at the polypeptide amide bonds. The resulting 'daughter' ion spectrum can be interpreted to give the peptide sequence. As the 'footprint' of mass spectrometers becomes smaller and biochemists more aware of their analytical power, mass detectors on HPLC and CZE systems will not be an uncommon sight! Another important and unexpected benefit of ESMS is that the width of the molecular ion 'window' appears to give a good idea of protein quality. As proteins age in vivo or in vitro, sulphur-containing and aromatic amino acids oxidize, asparagine deamidates and chemical adducts may form. These changes increase both mass and heterogeneity of the product. Comparison of the peak width for human serum albumin (Fig. lb) with that of the recombinant protein (Fig. la), shows
48+~
a
80
51 -t-
40 2 _
.
,
1100
.
1200
.
.
,
.
.
.
.
.
.
.
1300
.
,
b
10(
.
.
1400
.
.
1500
1600
1700
1800
1
6( 4( 2( (
.
1100
.
.
.
,
.
1200
.
.
.
.
.
.
.
.
1300
.
.
.
.
.
.
.
1400
,
.
1500 M/z
.
.
.
.
.
.
.
.
1600
.
.
.
.
,
.
.
1700
,
.
.
.
.
1800
ESMS spectrum of (a) recombinant (1-585), and (b) native human albumin (rHA and HSA). The spectrum shows M/z (mass : charge ratio) (n is equivalent to z in this instance) values between 1100 and 1860. Each peak represents a molecular ion Mn = [M ÷ nil+] n+ where M is the molecular weight, and n is the number of positive charges (in this case protons), n can be calculated from two adjacent ions, e.g. m7 and m2. Then n = (mT-l)/(mE-m~) and M = n(m2-1). In (a) two peaks are labelled (n+). The arrow indicates that HSA has a slightly greater average mass than rHA, and it is clearly more heterogeneous [compare peak widths for (a) and (b) at 10% peak height]. (The ESMS spectra were obtained using a BIO-Q MS, courtesy of VG Biotech Ltd. The rHA was provided by DELTA Biotechnology, Nottingham.)
that even when the individual mass differences cannot be resolved, the peak width at a set height may be used as a criterion of acceptance or rejection for a therapeutic protein. The reported higher quality of recombinant human protein products, as determined by various biophysical criteria, reflects their improved homogeneity compared with their naturally derived human counterparts. The desorption process in ESMS is still not well understood. The gaseous protein ions, freed of their solvent shells, probably unfold. This is because protein folding is an entropic process, driven by the burying of hydrophobic amino acids in the protein core away from the aqueous phase. Multiple charging probably helps the opening-up process through electrostatic repulsion. Bound ligands and cofactors (such as haeme groups) are stripped off and protein subunits are separated to leave linearized apoproteins. Covalent bonds are not broken in the process and it is noticeable that proteins (such as albumin) which are rich in disulphide bonds can accept fewer charges than smaller mol-
ecules which do not form disulphide bonds. This probably results from the inability of the crosslinked polypeptide chain in these proteins to extend fully. The relatively gentle ionization and desorption in ESMS offers considerable advantages over the older techniques of FAB and plasma desorption (PDMS) (see Glossary). For these ion sources, the very high energies needed to remove protein ions from liquid and solid matrices limits the methods to masses Of 10-20 kDa and the accumulation of data is slow (minutes to hours). In addition, both methods cause fragmentation of proteins which, coupled with the interference by matrix ions, limits both mass range and accuracy. This reduces the likelihood of being able to identify posttranslational modifications without first fragmenting the protein and separating the modified peptides. Since ESMS is able to measure both high and low molecular weight species accurately, further development of ESMS for protein sequence analysis is expected. The very valuable information now obtainable quickly and simply
303
TIBTECH- NOVEMBER 1990 [Vol. 8]
~Fig. 2 lOO 95 90 85 8o 75 70 65 60 55 5o 45 40 35 3o 25 20 15 lO 5 o~ 900
19+
charged with responsibility for scrutiny of each prospective recombinant therapeutic protein submission for a product licence. Their major concerns are for the safety, purity and, in many cases, nature identity of the product. No doubt protein mass spectrometry will now have to be added to the battery of assays and analyses currently undertaken by the biotechnology industry. For the research worker, however, the new technology is likely to prove both enabling and time-saving 5. Interesting results are certain to be reported as the use of ESMS becomes more widespread.
18+
0
0
1000
1100
1200
1300
1400
1500
M/z
ESMS of single domain rHA (1-194). There are at least two series of molecular ions. Series 0 differs from series R by 119Da. The expected mass is 22233.94Da, which differs little from that determined for the R series (22232.9_+ 1Da). The mass determined for the 0 series is 22352.3_+ 1.6Da [ 0 R = 119 Da]. Reduction o f the protein with dilute thiol causes series 0 to disappear and series R to increase in abundance. This can only result from cysteine having been picked up from the culture m e d i u m by formation of a disulphide bond to a free protein thiol. -
References 1 Meng, C. K., Mann, M. and Fenn, J. B. (1988) Z. Phys. Ser. D 10, 361-368 2 Dole, M. eta]. (1968) J. Chem. Phys. 49, 2240-2249 3 Mann, M., Meng, C. K. and Fenn, J. B. (1989) Anal. Chem. 61, 1702-1708 4 Morris, H. R. and Greer, F. M. (1988) Trends Biotechno]. 6, 140-147 5 Dorsselaer, A. et al. (October 1990) Biomed. Environ. Mass Spec. (in press)
MICHAELGEISOW for proteins of relatively high molecular weight should re-awaken the interest of biochemists and biotechnologists in mass spectrometry and send them hunting through their
notes for the masses of the elements. It should also stimulate the interest of the regulatory authorities, such as the Food and Drugs Administration (FDA). The latter, of course, is
Biodigm, 115 Main Street, East Bridgeford, N o t t i n g h a m NG13 8NH, UK.
